Abstract: Aboveground dry biomass is an important indicator for crop growth and yield. An experiment with different irrigation treatments was carried out to examine the changes of aboveground dry biomass of winter wheat. The predictive models for monitoring aboveground dry biomass were established using the methods of MLR and SPA-MLR. The results showed that irrigation at the jointing stage, booting stage, flowering stage and filling stages was a favorable method for aboveground biomass accumulation of winter wheat. The predictive models established with the method of SPA-MLR were more accurate than those built with the MLR method. Among them, the predictive model at the flowering stage of winter wheat performed best, with R²=0.96, RMSE=0.092 and R²=0.76, RMSE=0.18 for the calibration set and validation set, respectively. The predictive model covering all the growth stages from the jointing stage to filling stage of winter wheat achieved good prediction with R²=0.64, RMSE=0.30 and R²=0.54, RMSE=0.26 for the calibration set and validation set, respectively. Such a result indicated that this model could be potentially used to monitor the aboveground dry biomass in the winter wheat field as good predictive accuracy under extensive growth stages was achieved. Our results provide technical support for predicting the aboveground dry biomass of winter wheat using the hyperspectral technology.

Key words: dryland farming, precipitation simulation, soil CH₄ flux, short-term response.